Artificial intelligence and the intelligent edge

Micron Technology | January 2020

Will robots and holograms make us feel like we live in a science fiction world?

The science fiction future is one where robot droids work alongside humans, where holographic messages are beamed across space and where hyperdrive allows spaceships to travel at the speed of light. As 2020 heralds a new decade, we can claim two out of three as realities of today. Micron’s innovations in memory and deep learning acceleration are instrumental in making intelligent robots and virtual reality the transformational tools they have become. By 2030, according to Price Waterhouse Coopers, Artificial Intelligence (AI) I will add more than $15 trillion to the US economy.

The "intelligent edge" refers to the continually growing set of connected systems and devices where data is analyzed and aggregated close to where it is captured. Intelligent devices – those using AI for inference at the edge – are now more self-contained and do not need to connect to the cloud to run their algorithms. The application of AI is typically thought of in two parts:

  • Training that is done on a vast amount of data (usually in the cloud) to develop an algorithm to produce a desired result.
  • Inference, which means applying the algorithm on new data to put it to use.
For today’s intelligent edge devices, users expect a response in milliseconds, not seconds. This requires the device itself to do the inference and process the data in place, without sending data to the cloud for remote processing. Device makers design in FPGAs or GPUs to do faster neural processing, and they use a lot of DRAM to supply them with the data needed. As a result, they have the flexibility to tweak algorithms as part of the training process and can adjust the hardware configuration for optimum results.

Micron has a deep learning accelerator, comprising both a hardware board and a full SDK, that allows our customers to explore deep learning solutions required for data analytics, particularly in the internet of things (IoT) and edge computing. We have integrated compute, memory, tools and software into a comprehensive AI development platform, providing the key building blocks to develop innovative memory optimized for AI workloads.

Micron’s high-speed LPDDR5 DRAM and high-capacity SSDs are used to store and serve data for fast inference in edge devices, and our managed NAND and low-power DRAM solutions drive the intelligence in smartphones.

Micron is where the data lives. Micron — Intelligence Accelerated.